NonpartisanIndependent, data-first coverage of US elections — polls, forecasts, and prediction markets, refreshed daily.Updated daily.What's new·Methodology·Send feedback
Debrief · 2006 cycle
The 2006 Debrief · post-mortem · permanent archive

How did the polling industry do this cycle?

The 2006 industry MAE landed at 5.5pt — graded across 66 historical races with a known outcome — and 36 polls were analyzed across 23 pollsters. Signed bias −1.86pt: pollsters again systematically over-predicted Democrats.

INDUSTRY MAE
5.5pt
vs midterm avg
SIGNED BIAS
−1.86pt
D-leaning
POLLS · POLLSTERS
36
23 firms
RACES SCORED
66
w/ known outcome
CYCLE200220042006200820102012201420162018202020222024
Industry mean absolute error · 2002–2006
20024.0pt20044.5pt20065.5pt20084.3pt20105.2pt20124.7pt20144.8pt20166.8pt20184.5pt20206.0pt20224.6pt20244.1ptMAE · 2-WAY MARGIN
Per-pollster scorecard · 2006 · sorted by MAE
Signed bias · ±6.5pt
MAE · 0–10.4pt
rasmussen
−0.19pt
4.0pt
survey-usa
−4.35pt
4.3pt
zogbywsj
+6.05pt
9.9pt
Biggest-miss races · 2006
RaceExpectedActualMissDirection
us-governor-2006-nyD+45.00D+31.61-13.4ptD-margin overestimate
us-governor-2006-maD+25.00D+20.08-4.9ptD-margin overestimate
us-governor-2006-coD+12.00D+16.824.8ptD-margin underestimate
us-senate-2006-flD+26.00D+22.20-3.8ptD-margin overestimate
us-governor-2006-ohD+25.00D+23.89-1.1ptD-margin overestimate
us-governor-2006-orD+7.00D+7.981.0ptD-margin underestimate
us-senate-2006-meR+53.00R+53.42-0.4ptR-margin underestimate
us-senate-2006-nvR+14.00R+14.37-0.4ptR-margin underestimate